Radio operators transmit and, seconds later, hear their own signal return, far too late for any trip around the Earth. Observed since 1927, still reported by amateurs today, and still without an accepted mechanism in its long form.
Documented
Hals hears his signal twice
In 1927 Jorgen Hals, an engineer and amateur operator outside Oslo, heard the Dutch station PCJJ twice: the normal round-the-world echo at about a seventh of a second, and a weaker return about three seconds later. He wrote to physicist Carl Stormer, who put the puzzle in Nature in November 1928, with Balthasar van der Pol following a month later. On October 24, 1928, the echoes were heard simultaneously in Oslo and Eindhoven, which converted most skeptics. Reports never stopped: delays from under a second to thirty, on frequencies from 1.8 to 1296 MHz, logged from the 1920s to the present.
Assessment: The observations are documented in Nature and in a century of amateur logs. What is not documented is a mechanism for the multi-second kind: Shlionskiy's catalogue lists fifteen candidates, Vidmar and Crawford's 1985 review keeps five, and the one that is actually understood, magnetospheric ducting, tops out near half a second.
Stories
The ham radio ghost story
The echo lives in operator folklore: winter nights, low solar activity, the 1 to 4 MHz band, your own call sign coming home late. Along the way the phenomenon was absorbed into the Black Knight satellite legend, a conflation its most famous decoder never made and later explicitly disowned.
Assessment: The folklore and the Black Knight entanglement are both documented, and the entanglement is documented as false. The strip-mining of a genuine physics anomaly into a UFO franchise is itself the documented story here.
Interesting claims
The probe reading, published in Nature
In May 1960 Stanford's Ronald Bracewell argued in Communications from Superior Galactic Communities that an advanced civilization would pepper likely stars with autonomous probes, and that such a probe would announce itself by repeating our transmissions back: it would look exactly like echoes with delays of seconds or minutes, such as were reported thirty years ago by Stormer and van der Pol and never explained. In 1973 Duncan Lunan plotted the 1928 delay times, read a star map of Bootes out of them, and proposed a probe parked near the Moon for 13,000 years. In 1976, having found errors in his own method, he withdrew the reading.
Assessment: Both texts are documented: the probe hypothesis in Nature, and the retraction in Spaceflight. The phenomenon outlived both papers, which is the only claim in this entry nobody contests.
Sources and further reading
Stormer, Short wave echoes and the aurora borealis, Nature 122 (1928), 681; van der Pol, Nature 122 (1928), 878
Bracewell, Communications from Superior Galactic Communities, Nature 186 (1960), 670-671
Lunan, Space Probe from Epsilon Bootis?, Spaceflight (April 1973), and his retraction in Spaceflight (1976)
Vidmar and Crawford, Long-delayed radio echoes: mechanisms and observations, Journal of Geophysical Research (1985)
Holm, Long delayed radio echoes, the illusive secret of the ionosphere, HamSCI workshop (2021): five mechanisms, one understood
Worth asking
A Stanford professor told Nature the echoes could be a parked alien probe. The echoes continue a century later, and nobody has run his test. Why not?